Dr George Loizou
Health and Safety Laboratory
Buxton SK17 9JN
Telephone: +44 1298 218 438
The objective of this project is to further develop RVis, a prototype application for the analysis of structure and performance of physiologically based pharmacokinetic (PBPK), and other models, written in the free, open source syntax R or C++. The overall aim is to extend, improve and to provide more features and make them more robust.
The widespread adoption and application of PBPK modelling in product development and safety assessment has been hampered by criticism that these models are data hungry, resource intensive, complex and require high levels of mathematical expertise and programming skills. Most criticisms can be addressed, as has been demonstrated, with the development of prototype, proof-of-principle, user-friendly web-based tools such as MEGen (Loizou and Hogg, 2011), for the rapid generation of PBPK model code, and PopGen (McNally et al., 2014), a virtual human population generator. Both applications shift the emphasis away from the need for high levels of mathematical expertise and programming skills to the understanding of the biology of toxicity and disease that should underpin chemical safety and risk assessment. Further development of such tools would continue to mitigate existing concerns and make this powerful approach more readily accessible to safety toxicologists and risk assessors.
However, the greatest obstacle to the more widespread adoption of PBPK modelling is most likely the availability of a common, transparent and independently auditable, free-to-use platform for running models and analysing model structure and output. In response to this need the European Partnership for Alternative Approaches to Animal testing (EPAA) and the Health and Safety Executive (HSE) funded the Health and Safety Laboratory (HSL) to develop a user-friendly in vitro and in vivo exposure predictor. The motivation for this tool is the ultimate full replacement of animal testing which requires the ability to predict equivalent human oral, dermal or inhalation exposures that are consistent with measured in vitro target tissue concentrations; an issue which can only be achieved using PBPK modelling approaches. The output of this project was RVis, a prototype, proof-of-concept application for the analysis of structure and performance of PBPK, and other models, written in the free, open source syntax R or C++.
RVis is, in fact, a general purpose modelling platform, not just an in vitro and in vivo exposure predictor. The features of RVis include the ability to load, run, visualise and plot graphical outputs from models. Model structure may be analysed using parameter elementary effects screening and global sensitivity analysis (GSA) and parameter estimation using Markov Chain Monte Carlo simulation and Bayesian inference. The parameter estimation feature is used to perform “reverse dosimetry” to reconstruct human dose or exposure concentrations consistent with measured in vitro target tissue concentrations.
In response to the data security concerns of EPAA partners, RVis was designed to be installed on a user’s Windows- based PC or laptop thereby obviating the need for the uploading of (proprietary) data to a web-based application.
Four work packages are proposed for the further development of RVis. The work packages are based on the recommendations made by the independent evaluation of the prototype of RVis organised by ECETOC on behalf of CEFIC. The evaluations were conducted by experts from the European Chemicals Agency (ECHA), EU Joint Research Centre (Italy), US Environmental Protection Agency, US Food and Drug Administration, Texas A&M University, Sumitomo Chemical, Shell (The Netherlands), Fraunhofer ITEM (Germany) and Wageningen University (The Netherlands).
The work packages represent themes for the overall next stage of development of RVis. The relevant tasks required to deliver the theme are grouped within each work package and correspond to the work flow described in RVis i.e., load and run model, initial investigation of model by selecting and tweaking parameters followed by elementary screening prior to more computationally intensive global sensitivity analysis and finally parameter estimation . They do not conform to a chronological sequence of tasks required to deliver the final product where any given task is dependent upon completion of the previously listed task. The chronological sequence of tasks comprising the research plan is presented in the section for Specific milestones, decision points and timing.
Work package 1: Improvements to usability
Work Package 2: Improvements and extensions to sensitivity analysis feature
Work Package 3: New feature for batch processing operations
Work Package 4: Improved parameter estimation feature